BloodAxe/workshop-gan-2018

Plan

1. Intro
2. Setup
3. GAN trainer
4. Class imbalance

Setup

virtualenv -p python3.6 venv
source ./venv/bin/activate
pip install -r requirements.txt
export PYTHONPATH=.

load checkpoint and dataset here

Run tensorboard

tensorboard --logdir . --port 8887 --host 0.0.0.0

Class Imbalance

./run.sh

MNIST GAN trainer

python -m experiments.train_gan  --config-path ./configs/train_gan.json

MNIST conditional GAN trainer

python -m experiments.train_cgan  --config-path ./configs/train_cgan.json

References

• Generative Adversarial Networks
• A systematic study of the class imbalance problem in convolutional neural networks
• BAGAN: Data Augmentation with Balancing GAN
• Deep Imbalanced Learning for Face Recognition and Attribute Prediction
• A Classification–Based Study of Covariate Shift in GAN Distributions
• Which Training Methods for GANs do actually Converge?
• An empirical study on evaluation metrics of generative adversarial networks
• Improved Techniques for Training GANs
• Medical Image Synthesis for Data Augmentation and Anonymization using Generative Adversarial Networks
• Data Augmentation Generative Adversarial Networks